Development of a control model for predictive control data to extend the process limits of piezokinematic systems for production technology.

该项目的目的是开发一种用于机床的高动态压电轴的预测控制模型。借助于要开发的控制模型，应该可以使用在允许的控制器带宽之上的附加轴。应该使用控制模型来准确地预测位置，而不必经历复杂的迭代过程。这应该有助于扩大现有机床系统的生产限制，并增加压电系统的灵活性和性能。建模应基于已制造工件的触觉测量几何数据。通过使用工件测量数据，刀具和工件的动态影响将包含在模型中。为了从工件测量数据进行建模，必须对合适的系统辨识方法进行比较和评估。鉴于目前的技术发展，应该把重点放在用神经网络进行机器学习的方法上。为了进行建模，必须将测量的工件数据转换到时间域、频率域或拉普拉斯域，以提取合适的特征。模型的带宽由所选的用于工件加工的测试几何确定。为了增加模型带宽，必须改变试验几何的类型、加工速度、几何在工件中的角度对齐和刀具的偏转方式。具有附加压电轴的三轴加工中心可用于工艺条件下的测试和验证。保证了机器行为的必要的重复性或时间不变性。

The aim of the project is to develop a predictive control model for highly dynamic piezo axes for use in machine tools. With the aid of the control model to be developed, it should be possible to use the additional axes above the permissible controller bandwith. The control model should be used to predict the position accuratly without having to go through a complex iterative process. This should help to extend the production limits of existing machine tool systems and to increase the flexibility and performance of piezo systems. The modeling shall be based on tactile measured geometry data of manufactured workpieces. By using the workpiece measurement data, dynamic influences from the tool and the workpiece are to be included into the model. For modeling from the workpiece measurement data, suitable methods of system identification must be compared and evaluated with each other. In view of current technological developments, one focus should be put on methods of machine learning with neural networks. For modeling purposes, the measured workpiece data must be transformed into the time, frequency or laplace domain to extract suitable characteristics. The bandwidth of the model is determined by the selected test geometries used for the workpiece machining. In order to increase the model bandwidth, the type of test geometries, the machining speed, the angular alignment of the geometry in the workpiece and the deflection mode of the cutting edge must be varied. A 3-axis machining centre with additional piezo axes is available for teh and verification under process conditions. The necessary reproducibility or time invariance of the machine behaviour is guaranteed.